Posted by killbatman on 7/25/2012 2:29:00 PM (view original):
I wanted to elaborate on a point red made about stats vs. ratings. I'll go almost completely by ratings at the start of the season..I might consider past seasons' stats a little, but not much. Then maybe about 5-6 games into the season, I'll start monitoring the stats and giving them more and more weight in my mind. Call it superstitious if you want, but I definitely watch for streaks in this game (e.g. the infamous SR slump). I'm much more likely to bench or lower the distro on an underperforming player than to just sit back and hope he turns it around.
Lineups seem to "click" in a way I don't understand sometimes, so I think it's important to watch the results. I'm a big believer in tweaking the distro throughout the season and eventually settling into a "best lineup" before the postseason starts.
this is a pretty controversial issue in HD - do you set distro/depth chart by ratings or by stats? i guess the logical answer, in many peoples minds is, just play by the ratings, that is all that goes into the game (and settings). and i see the merit in it. plus, early on, stats mean SO LITTLE. you might have a guard shoot 6 shots a game, a decent number, and after 7 games hes got 42 attempts. its really easy on that small of a sample for him to miss a few extra - just making or missing 2 extra baskets at that number of attempts has a 10% swing! (2 extra makes over expectation is 5% higher). and thats a VERY VERY moderate variation on 42 attempts. so you might have a 45% shooter shooting 35% or 55% and really that is not a big deal, on that sample size.
i personally am a coach who makes changes based on actual outcomes, but i usually wait at least like 10-14 games (and at 10 i usually still won't tweak more than maybe 1 because i still want to extra data). my reasoning is this - none of us are perfect. and nobody knows WTF is going on inside that sim engine :) so you might be a good coach and think your 70 ath 70 spd 80 per guard might be better than your 40 ath 85 spd 85 per guard, so you give him more distro. but maybe hes under performing, because you were incorrect in your assumption! after a couple hundred attempts, its hard to write it ALL off as luck, so i do start to adjust.
the other part i mentioned is, nobody REALLY knows what the sim engine is doing. you might have a better scorer who is getting penalized by fighting the best defender, a slightly worse scorer getting penalized fighting also a really good defender, but your 3rd scorer might be playing backup or sf or something, and wind up with easier competition. so, he might pull the best stats on the team, and he might do so even if you ran the season a million times. plus, there are definitely times the sim engine just does some weird stuff. so i think its unrealistic to think there is nothing to be gained by looking at the actual outcome. YOU JUST REALLY HAVE TO BE CAREFUL OF SAMPLE SIZE!! especially on fg%. you can do it on mpg and stuff a lot easier if you check the box score carefully for fouls etc.
in the end of the day, if i am highly confident i understand the abilities of a player who is under perfoming, i still generally stick with my gut, ratings wise. i mean often ppl go damn, that guard i expect to be 45/40% is 40/35%. you know how likely that is in a simple bell curve? quite likely. so you can kind of turn to stats here and there, especially the less certain you are about what you are doing. but when you are confident in your understanding of the mechanics and ratings, you have to 80-90% rely on what should work and 10-20% on the actual results.